What is GPT doing

By Stephen Wolfram

This is a short book which combines an description of how GPT, the revolutionary new AI technology shaking the world, works with an advertisement for some Wolfram products designed to augment GPT.  His description is pretty good but I couldn’t really follow his explanation of the crucial “transformer” component, the attention blocks.  ChatGPT was able to give me a better account.

Wolfram pointed out some key failings of GPT.  It can’t really do deductive chains of any substance.  For example, ask it to compute 3^73 and it comes up with an absurdly wrong answer, off by a factor of roughly 10^34.  I tried giving it some hints, including using logarithms.  Again, it confected a plausible looking deductive chain and an answer only off by a factor of ten or so.

Wolfram then argues that supplementing GPT with some deductive system like Wolfram language would have more capability.  Of course lots of people are working on such things.

Comments

Popular posts from this blog

Anti-Libertarian: re-post

Uneasy Lies The Head

We Call it Soccer